基于改进差分进化算法的非线性系统辨识

Identification of Hammerstein Model Based on Improved Differential Evolution Algorithm

在工业控制系统非线性优化问题的研究中,关于典型非线性系统的参数辨识问题,为了避免传统辨识方法适用性狭小、鲁棒性差等缺点,引入基于种群的启发式算法,提出利用标准差分进化算法的非线性系统模型辨识方法。为了进一步克服启发式算法的早熟收敛问题,采用在模糊逻辑控制下的参数自适应差分进化算法,证明了自适应的参数选择能够降低参数对实际问题的依赖性。通过实例仿真,结果与标准差分进化算法相比较,表明了改进算法的可行性和有效性。

For the identification of nonlinear system, we put up an normal differential evolution （DE） algorithm. In order to avoid the performance of precocious effectively, a parameter adaptive DE algorithm controled by a fuzzy logic was also presented. Compared with the original DE Algorithm in numerical simulations, the present method was verified to be practical and effective.